With the rapid growth of digital content and the increasing demand for high-resolution displays, the efficient compression of screen content images characterized by text, graphics, and UI elements has become an important research field. This paper introduces a new dataset named SCID-Compress900 specially designed for image compression research. The dataset consists of 900 high-quality screen content images, including 500 4K images and 400 1080P images. All these images are mainly composed of text/graphics, reflecting typical screen content scenarios such as office documents, software interfaces, and presentation slides. The dataset covers a diverse range of content, including various font sizes, graphic styles, and color modes, providing a comprehensive testbed for compression algorithms. To demonstrate the effectiveness of SCID-Compress900, we conduct benchmark tests using several deep learning-based image compression methods commonly employed by researchers. The experimental results show that SCID-Compress900 can well differentiate the performance of different compression algorithms. Compared with existing datasets, SCID-Compress900 offers higher resolution, larger scale, and more targeted content, making it an ideal resource for developing and evaluating advanced image compression algorithms for screen content. This dataset will not only promote the research and development of screen content compression technology but also contribute to the standardization and optimization of compression algorithms in practical applications. The project is available at https://openi.pcl.ac.cn/OpenDatasets/SCID-Compress900.

Limited by the sensor resolution of plenoptic cameras, it is challenging to obtain spatial and angular high resolution light field (LF). In this paper, we propose a method for spatial and angular super-resolution (SR) simultaneously while preserving parallax structure, which generates a high resolution (HR) dense LF from a low resolution (LR) sparse one with detailed textures. The proposed framework consists of four parts, including a depth estimation module to get scene geometry, an image warping process to generate intermediate warped views, an epipolar plane image (EPI) generation module to recover structural features, and a spatial-angular SR module to fully explore the informative features in 4D LF. Finally, the high quality HR scenes can be effectively reconstructed. Experimental results over several LF datasets demonstrate that our approach not only outperforms the existing spatial and angular SR methods separately, but also achieves better quantitative and qualitative results than the one-stage and two-stage spatial-angular approaches.

With the development of Connected Autonomous Vehicle (CAV) technology, different information flow topologies (IFTs) have been applied to CAV Ad Hoc Networks. Firstly, from the perspective of the controller, a general model is proposed to directly reflect the actual communication effect on the controller instead of simply abstracting it into the optimal time interval, which is more feasible. Secondly, linear stability analysis is carried out based on the general model where different time delays are considered, and stability criterion is obtained for subsequent analysis. Finally, we compare the three main IFTs through numerical simulations and analyze the difference in stability region, robustness, traffic safety, and Eco-driving between platoon Cooperative Adaptive Cruise Control (CACC) controllers based on different IFTs. It is found that adopting CACCs can notably improve the traffic capacity and traffic safety relative to Manual Vehicle (MV) no matter what IFT is adopted. As for the difference between the three IFTs, predecessor-leader following (PLF) and multiple-predecessor-leader following (MPLF) are significantly superior to predecessor following (PF), while the choice between PLF and MPLF depends on the communication bandwidth. With higher communication bandwidth or fewer communication vehicles, MPLF will be a better option; on the contrary, PLF is more suitable.

View synthesis (VS) for light field images is a very time-consuming task due to the great quantity of involved pixels and intensive computations, which may prevent it from the practical three-dimensional real-time systems. In this article, we propose an acceleration approach for deep learning-based light field view synthesis, which can significantly reduce calculations by using compact-resolution (CR) representation and super-resolution (SR) techniques, as well as light-weight neural networks. The proposed architecture has three cascaded neural networks, including a CR network to generate the compact representation for original input views, a VS network to synthesize new views from down-scaled compact views, and a SR network to reconstruct high-quality views with full resolution. All these networks are jointly trained with the integrated losses of CR, VS, and SR networks. Moreover, due to the redundancy of deep neural networks, we use the efficient light-weight strategy to prune filters for simplification and inference acceleration. Experimental results demonstrate that the proposed method can greatly reduce the processing time and become much more computationally efficient with competitive image quality.